Effectiveness of Ignition Interlocks forPreventing Alcohol-Impaired Driving and
Alcohol-Related CrashesA Community Guide Systematic Review
Randy W. Elder, PhD, Robert Voas, PhD, Doug Beirness, PhD, Ruth A. Shults, PhD, MPH,David A. Sleet, PhD, FAAHB, James L. Nichols, PhD, Richard Compton, PhD,
From the Division of Community Preventive Services, Epidemiology andAnalysis Program Offıce (Elder), the Alcohol, Policy, and Safety ResearchCenter, Division of Unintentional Injury Prevention, National Center forInjury Prevention and Control (Shults, Sleet), CDC, Atlanta, Georgia; Pa-cifıc Institute for Research and Evaluation (Voas), Calverton, Maryland;Beirness &Associates Inc. (Beirness), Nepean, Ontario, Canada; andOffıceof Behavioral Safety Research, National Highway Traffıc Safety Adminis-tration (Compton, Nichols), Washington DC
Names and affıliations of Task Force members are available atwww.thecommunityguide.org/about/task-force-members.html.
Address correspondence to: Randy W. Elder, PhD, Guide to Commu-nity Preventive Services, Epidemiology andAnalysis ProgramOffıce, CDC,1600 Clifton Road, Mailstop E-69, Atlanta GA 30333. E-mail: [email protected].
Drivers convicted of driving while impaired(DWI) present a high risk to other highwayusers. Hedlund and Fell1 found that offenders
convicted of DWI are 4.1 times more likely to be in-volved in a fatal crash while intoxicated by alcohol thanare average licensed drivers. Further, 35% to 40% of allfatally injured drinking drivers are estimated to havehad at least one prior DWI offense.2,3 For the fırst twothirds of the 20th century, the traditional penaltiesassessed for a DWI conviction were jail, fınes, andlicense suspension. Of these, license suspension hasprovided the strongest andmost consistent evidence ofeffectiveness in reducing recidivism.4–6 Nonetheless,oth self-reports7 and covert surveillance8 of sus-ended DWI offenders indicate that many of these
rivers continue to drive without licenses or insurance,
dbyElsevier Inc. on behalf ofAmerican Journal of PreventiveMedicine
Elder et al / Am J Prev Med 2011;40(3):362–376 363
M
and that they often continue to do so even after becom-ing eligible for license reinstatement.9
Use of Vehicle Sanctions to Reduce IllicitDrivingGiven the evidence that suspension alone does not pre-vent DWI offenders from driving illicitly, state legisla-tures have enacted various measures that keep impaired-driving offenders from accessing their vehicles, such asvehicle impoundment and immobilization laws10–12 andehicle forfeiture laws.13 Although there is evidence thatsuch actions reduce recidivism, they may have potentialadverse effects on innocent family members, and there-fore have not been widely adopted for a broad range ofoffenders. A less disruptive approach to reduce DWIrecidivism is to require installation of a device on thevehicle that prevents a driver impaired by alcohol fromoperating the vehicle.
History of Interlock Development and UseThe fırst efforts to develop devices to prevent vehicleoperation by intoxicated drivers grew out of human per-formance research. From this work, some prototype de-vices, such as a “Quick Key” unit that tested the driver’sreaction time, were developed for vehicles.14 However,the large variation in human performance resulted in asubstantial number of false-positive signals. Subsequentsystems were designed to assess intoxication more di-rectly, using a dashboard-mounted breathalyzer devicethat prevents the vehicle from starting if the driver’sblood alcohol concentration (BAC) is above a predefınedlimit. The fırst ignition interlock device that used BACtesting was introduced in 1970. However, it did not oper-ate reliably under all environmental conditions, andlacked several provisions that proved necessary to pre-vent circumvention.Beginning in the 1990s, a “second generation” of inter-
locks15 was introduced, which added several features thatmade circumvention more diffıcult. These featuresinclude:
● Hum tone recognition—which requires training tomake the sensor work, and prevents the driver fromusing untrained substitutes,
● Filtered air detection—which prevents blowingthrough a device that fılters out the alcohol,
● Blow abort—which detects air samples that are toosmall, and
● Random running retest—which prevents drinkingwhile the engine is running.
Coupling these features with a requirement that the
interlocked vehicle be brought in for service every 30 days
arch 2011
resulted in units forwhich undetected circumventionwasvery diffıcult.16 These features were integrated into aodel for state interlock standards by theNational High-ay Traffıc Safety Administration (NHTSA) in 1992. Al-hough these standards minimized the opportunity forffenders to drive their interlock-equipped cars afterrinking, they could not prevent offenders from circum-enting the system by illegally driving different vehiclesithout interlocks. However, the existing evidence17 sug-
gests that the availability of a non-interlock vehicle didnot greatly reduce the effectiveness of interlocks for pre-venting alcohol-impaired driving.The advent of “second generation” interlock units and
the 1992 NHTSA standards stimulated a dramatic in-crease in their use in the U.S. (Figure 1), amounting to anestimated 200,000 units by 2009.18 Nevertheless, NHTSA’sodel specifıcations still showed two limitations in the
nterlock hardware: (1) defınitive identifıcation of theerson blowing into the unit was not available and (2)lthough circumvention was diffıcult, detection of anyttempt to circumvent the unit was delayed until the nextcheduled maintenance inspection of the interlock. Sincehen, interlock manufacturers have been working to im-rove methods for identifying the user and for ensuringhat an attempt to bypass the unit will cause it to shutown unless it is quickly brought in for a maintenancenspection (�48 hours).
Types of Interlock Programs and TheirInfluence on Interlock UseOne of the most important limiting factors for the publichealth impact of interlock programs is the relatively smallnumber of offenders who participate in such programs.Despite the continuing growth of interlock use, only asmall fraction of the approximately 1.4 million peoplearrested for DWI annually in the U.S. use them. Thisapparently low usage rate is partly due to the failure to
Figure 1. Interlock growth in the U.S. through 2009, datawith polynomial fit (adapted from Marques, 200918)
convict some of the arrested offenders, and it also reflects
sapfe
eqf
364 Elder et al / Am J Prev Med 2011;40(3):362–376
the policy in most states of confıning their programs tothe 400,000 to 500,000 multiple offenders apprehendedeach year. Furthermore, contrary to early expectations,many DWI offenders prefer license suspension to inter-lock installation; generally, less than 10% of eligible of-fenders enter interlock programs.19
The specifıc features and procedures of interlock pro-grams vary both among states and among jurisdictionswithin states. One of the key features of interlock pro-grams in the U.S. is the type of legal authority that theyoperate under:
1. The judicial authority to impose suitable sanctions un-der common law;
2. State laws specifıcally providing for the use of inter-locks as a sanction for DWI or for driving while sus-pended (DWS); and
3. The administrative authority of the state motor vehicledepartment to regulate driver licensing.These programs also vary with respect to who is eligi-
ble, when interlocks are installed (i.e., after arrest versusafter sentencing), and whether judges or offenders havediscretion over offering or participating in interlock pro-grams. All of these features have the potential to influenceboth the number and characteristics of offenders whoparticipate in the programs, and thus their public healthimpact.
Judicial Interlock ProgramsThe earliest interlock programs were implemented byindividual judges applying the interlock as one sanction-ing option. Under such programs, court orders to installinterlocks often were not processed by the department ofmotor vehicles (DMV). Consequently, these requiredsanctions were not on the license records, and offıcerswho stopped offenders would not be alerted to check forthe interlocks.Based on some evidence that these early programs
were successful in reducing recidivism among the smallnumber of people who had interlocks installed,20 sometates passed legislation that provided judgeswith explicituthority to impose interlocks at their discretion. Thisroduced only a modest increase in the number of of-enders using interlocks because the courts continued tonlist only 10% or less of eligible offenders.21
In part, this low penetration of interlock programsstimulated the passage of mandatory laws that requiredjudges to place offenders convicted of multiple DWIs ininterlock programs. However, such laws conflicted withother state legislation mandating hard license suspensions—which prohibit any driving—for second-time offenders.Thus, courts were mandated to require interlocks on
vehicles that offenders could not legally drive; few courts
complied with this mandate.22,23 A provision of the Fed-ral Transportation Equity Act for the 21st Century re-uired states to suspend the licenses of second-time of-enders for at least 1 year.24 That law has since beenchanged by the Congress to require only 3 months ofsuspension.
Administrative Interlock ProgramsSeveral states (e.g., California, West Virginia) have en-acted laws administered by stateDMVs that allowoffend-ers who have interlocks installed to drive during a periodin which their driving licenses would otherwise be fullysuspended. More recently, in part because of the require-ments of the Federal Transportation Equity Act, stateshave begun to implement laws that require installation ofan interlock as a prerequisite for license reinstatement.Two types of such legislation have been implemented.Some states require a period of interlock installation be-fore license reinstatement, but the offender can delayreinstatement during the period the interlock would berequired and thereby avoid installing the device. An alter-native implemented in other states, such as Florida, is torequire a period of interlock installation no matter howlong the offender delays, thus making it impossible toavoid the requirement if the offender is ever to be reli-censed. A potential problem for these postreinstatementprograms is that many DWI offenders delay reinstate-ment for 1 year or more and up to one third neverreinstate.9,25
Efforts to Increase Interlock UseSeveral recent efforts have been made to develop pro-grams that better address barriers limiting the uptake ofinterlocks. The goal of these efforts is to increase thenumber of offenders who drive interlock-equipped vehi-cles, to realize more fully the potential population-leveleffects of interlocks on alcohol-impaired driving andalcohol-related crashes. Primary strategies involve in-creasing the number of offenders eligible for interlocks,increasing the desirability of participating in the pro-gram, and increasing the negative consequences of failingto participate in the program.Some states have developed programs that allow or
require the installation of interlocks after an arrest (asopposed to conviction) for DWI. This minimizes theperiod of license suspension, duringwhich offendersmaybecome accustomed to driving illegally,8 thus leadingthem to devalue the opportunity for legal driving af-forded by the interlock program. For example, Texas hasa judicial program that requires some offenders to havean interlock installed before posting bail after arrest. In2003, New Mexico created a voluntary administrative
program that allows any driver whose license is sus-
www.ajpm-online.net
e
mufe
pbmr(
ofe
Elder et al / Am J Prev Med 2011;40(3):362–376 365
M
pended for an alcohol offense to receive a license to op-erate an interlock-equipped vehicle simply by appearingat the DMV with an interlock-equipped vehicle.26
New Mexico has made numerous other changes in itsinterlock program to increase its penetration. In 2003, thejudicial discretionary law was replaced by one that man-dated interlocks for all fırst-time offenders convicted ofaggravatedDWI and allmultiple offenders. In 2006, a lawmandating interlocks for all DWI offenders was imple-mented, greatly expanding the number of offenders eligi-ble for the program. This law includes provisions forsubsidized installation and monitoring of interlocks forlow-income offenders. Although implementation of eachlaw appears to have produced an increase in the numberof interlocks in use, by 2006, only 25% of New Mexicodrivers apprehended for a DWI offense had been placedon the interlock.The modest success of the efforts to increase the pene-
tration of interlocks in NewMexico illustrates the limita-tions of such efforts. Some of these limitations are inher-ent in the judicial system itself. In New Mexico, roughlyone third of those arrested for DWI are not convicted ofthat offense. Another problem relates to loopholes in thelegislation that allow offenders who claim not to havevehicles or who agree not to drive to avoid interlockinstallation. Those who use this loophole with the inten-tion of driving while suspended can then simply drive anunregistered vehicle or one registered to another person.This “no-car problem”23 is perhaps the major barrier toxtending interlocks to all convicted offenders.Efforts continue to develop interlock programs thatinimize the opportunities for offenders to avoid theirse. One such approach is to applymore-severe sanctionsor people who opt out of an interlock program. Forxample, one study19 found that if electronically moni-tored home arrest was presented as an alternative to in-terlock installation, approximately two thirds of DWIoffenders chose to have interlocks installed.
Goals of This ReviewOne of the primary research issues of interest to thereview development team was the degree to which theinstallation of interlocks in offenders’ vehicles reducesalcohol-impaired driving in comparison to alternativesanctions (primarily license suspension). Because thisquestionwas addressed thoroughly by a recent systematicreview conducted for the Cochrane Collaboration,27 therevious work was not replicated. The current reviewuilds on the earlier Cochrane review by incorporatingore recent studies and placing the fındings from thateview in a broader public health context. Specifıcally, it
1) addresses the effects of ignition interlock installation
arch 2011
n motor vehicle crashes and (2) discusses various keyeatures of interlock programs that could increase theirffectiveness for improving population health.
MethodsThis review was conceptualized and conducted by a systematicreview development team consisting of subject matter experts intraffıc safety and systematic review methodology, under the over-sight of the independent, nonfederal Task Force on CommunityPreventive Services (the Task Force) and using the methods of theGuide to Community Preventive Services (Community Guide).Community Guidemethods for conducting systematic reviews andlinking evidence to effectiveness are described in print elsewhere28
and on the Community Guide website (www.thecommunityguide.org/about/methods.html). In brief, for each Community Guidereview topic, a systematic review development team representingdiverse disciplines, backgrounds, and work settings conducts areview by (1) developing a conceptual approach to identify, orga-nize, group, and select interventions for review; (2) developing ananalytic framework depicting interrelationships among interven-tions, populations, and outcomes; (3) systematically searching forand retrieving evidence; (4) assessing and summarizing the qualityand strength of the body of evidence of effectiveness; (5) translatingevidence of effectiveness into recommendations; (6) summarizingdata about applicability (i.e., the extent to which available effective-ness datamight apply to diverse population segments and settings),economic impact, and barriers to implementation; and (7) identi-fying and summarizing research gaps. All data abstraction andquality scoring is conducted by two independent reviewers.
Conceptual Model
Figure 2 illustrates the hypothesized causal pathway from the im-plementation of ignition interlock programs through the outcomesof interest for judging effectiveness, specifıcally alcohol-impaireddriving and its consequences—alcohol-related crashes and the re-sulting fatal and nonfatal injuries. Solid lines reflect relationshipsassessed in this review, and dotted ones reflect those that are simplyhypothesized. It should be noted that although alcohol-impaireddriving (i.e., recidivism) was a primary variable of interest in thisreview, it can be assessed by only the proxymeasure of re-arrest foralcohol-impaired driving (or for related infractions).The initial step in this causal pathway is that ignition interlock
programs will result in the installation of interlocks in offenders’vehicles. As discussed above, the strength of that causal relation-ship is likely to be a function of the characteristics of the interlockprograms. If used properly, interlocks will prevent alcohol-impaired driving and its consequences. Installation of interlocksmay also have two secondary effects with important public healthconsequences. First, because they allow legal driving, they can beexpected to increase the number of miles driven by participants ininterlock programs relative to offenders who have had their li-censes suspended, potentially increasing crashes that are notalcohol-related, which may result in injuries. Second, by forcingparticipants to choose between drinking and driving, interlocksprovide a consistent behavioral consequence that may discouragedrinking in the short term. Interlocks also have substantial poten-tial for synergistic use with programs to address offenders’ under-lying alcohol dependence and abuse problems. These potential in-
fluences on alcohol consumption could contribute to reductions in
366 Elder et al / Am J Prev Med 2011;40(3):362–376
alcohol-impaired drivingand its consequences inboth the short and longterm.
Search for Evidence
In the interest of effı-cient use of resour-ces, Community Guidemethods allow for theincorporation of sys-tematic reviews con-ducted by other groupsinto the body of evi-dence that is used to as-sess the effectiveness ofinterventions. The pri-mary criteria that mustbe met to incorporatesuch reviews are thatthey conceptualize theintervention of interestin a similar manner to that of the review team; conduct a thoroughand clearly systematic literature search; and provide results in aformat and level of detail that adequately addresses one or more ofthe key research questions identifıed by the review team. In suchcases, supplementary searches for and evaluations of evidencemaybe conducted to answer other issues of importance to the reviewteam and the Task Force on Community Preventive Services.One of the primary research questions identifıed by the teamwas
the degree to which the installation of interlocks in offenders’vehicles reduces alcohol-impaired driving compared with alterna-tive sanctions. Because an initial search identifıed a recent system-atic review27 conducted for the Cochrane Collaboration that mettheCommunity Guide criteria for incorporating an existing review,no further database search was conducted for this review. Instead,a focused systematic search of sources likely to provide a high yieldof relevant studies was conducted to obtain information to fıll thegaps in the Cochrane review. The results from theCochrane reviewwere considered along with information from subsequent studiesthat were identifıed by hand-searching three key outlets for re-search on the effectiveness of ignition interlocks—the journalTraf-fıc Injury Prevention; publications of the International Council onAlcohol, Drugs, and Traffıc Safety’sWorking Group on Alcohol Igni-tion Interlocks; and the proceedings of the International Symposia onIgnition Interlocks. These hand-searches covered the period of March2003 through December 2007. Inclusion criteria for papers identi-fıed in the hand-search were identical to those used in the previousreview, with the exception that studies evaluating motor vehiclecrashes were also eligible for inclusion.Information on motor vehicle crashes and other variables of
interest (e.g., participation rate in the interlock program; programeligibility criteria)was abstracted from the papers identifıed in boththe Cochrane review and those subsequently identifıed in thehand-search. Inclusion criteria for papers identifıed in the hand-search were identical to those used in the previous review, with theexception that studies evaluating motor vehicle crashes were alsoeligible for inclusion. Four unique studies in addition to those from
Figure 2. Conceptual modelalcohol-impaired driving and itsrounded corners indicate mediaoutcomes)
the Cochrane review were identifıed in the hand-search, and one
tudy from the Cochrane review included information on motorehicle crashes.
Effect Measurement and Synthesis of Results
The majority of the studies included in this review reportedintervention effects as risk ratios (RRs, usually derived from a2 � 2 contingency table of events occurring at any time over thentire evaluation period) or as hazard ratios (HRs, usually de-ived from survival analyses). Because HRs address individualifferences in “time at risk” due to factors such as when inter-ocks were installed or removed, and whether offenders weree-arrested during the study period, HRs were reported whenossible. Both of these metrics can be similarly interpreted aseflecting the percentage change in risk of re-arrest attributableo the interlock intervention. Thus, to facilitate synthesis acrosstudies, these effect metrics were treated as equivalent for theurpose of calculating summary statistics. The results from theochrane review are summarized with descriptive statistics,pecifıcally medians and interquartile intervals. The additionalesults from papers identifıed in the hand-search are presentedeparately.
Results. Part I. Intervention and StudyCharacteristicsAppendix A presents the key intervention and studycharacteristics for the studies included in this re-view,21,23,29–43 as well as summaries of their fındings.Although several program characteristics variedwidely across studies (e.g., the type of program), all ormost programs shared several key characteristics. Theygenerally
● were applied to offenders who are at high risk of recid-ivism, either due to multiple offences or, for fırst-timeoffenders, high blood alcohol concentration (BAC) at
the causal effects of ignition interlock programs onsequences. (Ovals indicate interventions; rectangles withor intermediate outcomes; and rectangle indicates health
forcontors
arrest (generally �0.15 g/dL);
www.ajpm-online.net
m
esa
hafwrCrgs
apavr
nrrattn
Elder et al / Am J Prev Med 2011;40(3):362–376 367
M
● were offered as an option in exchange for a reducedlength of license suspension (and compared with driv-ers with suspended licenses as a comparison); and
● required interlock installation for periods rangingfrom 5 to 36 months (medians of 7.5 months for fırst-time offenders and 18 months for repeat offenders).Reported participation rates varied dramatically
across the programs evaluated in the reviewed studies,from less than 1% of offenders39 to 64% of offenders30
(median: 13%). This large variability partly reflects theuse of different denominators (e.g., interlock-eligibleoffenders versus offenders who actually were offeredinterlocks).The majority of reviewed studies prospectively fol-
lowed cohorts of offenders who had interlocks installedin their cars, and compared them to cohorts of offenderswho did not have interlocks and whose licenses weresuspended instead. Because several nonrandom factorscan influence whether a given offender has an interlockinstalled, such studies have a substantial risk of producingbiased effect estimates resulting from noncomparable in-tervention and comparison groups. Several studies in-cluded evidence suggesting at least some degree ofnoncomparability between groups. In particular, the in-terlock groups tended to be older,31,35,39 drive more,32
have higher incomes,33,39 and have more offenses orore serious offenses.33,41
Results. Part II. Evidence of EffectivenessEffects of Ignition Interlock Installation onDWI Recidivism
Results from the studies in the Cochrane review. TheCochrane review27 identifıed 11 studies that evaluated theffects of either the opportunity to have an interlock in-talled29,30 or of actual interlock installation31–39 on re-rrest rates for alcohol-impaired driving.The fırst study29 of the effects of the opportunity toave an interlock installed found that randomization ton interlock program through which 64% of eligible of-enders’ vehicles had interlocks installed was associatedith a 64% relative decrease in re-arrests during the pe-iod when interlocks were installed (RR�0.36, 95%I�0.21, 0.63); following the interlock installation pe-iod, re-arrest rates for the intervention and controlroups were similar (RR�1.33, 95% CI�0.72, 2.46). Theecond30 assessed changes in re-arrest rates for a countyin which the judge implemented an aggressive manda-tory interlock program (in which 62% of eligible offend-ers’ vehicles had interlocks installed) to those for sur-rounding counties. The authors found a 40% relativedecrease in re-arrest rates for fırst-time offenders
(p�0.04), and a 22% relative decrease for repeat offenders
arch 2011
(p�0.03) over a follow-up period that extended for sev-eral years beyond the removal of the interlock for somerepeat offenders (i.e., those whose DWIs occurred earlyin the study period).The nine studies31–39 that assessed the effects of inter-
lock installation consistently found that offenders whohad interlocks installed in their cars had recidivism rates(i.e., re-arrests) that were dramatically lower than driverswho did not have interlocks installed (median RR�0.25,interquartile interval [IQI]�0.18–0.46; see Appendix Afor study summaries). Effect estimates were similar forfırst-time offenders versus repeat offenders. For the pe-riod after these interlocks were removed, recidivism ratesin the intervention group tended to converge withthose for the comparison group (median RR�0.93,IQI�0.67–1.36).
Results from additional studies. The four identifıedstudies23,41–43 published subsequent to the Cochrane re-view found results consistent with those described above.Two studies evaluating the effectiveness of differentstages of New Mexico’s program found that interlockinstallation was associated with a 65% lower risk of recid-ivism among repeat offenders (HR�0.35, p�0.01),43 and61% lower risk among fırst-time offenders (HR�0.39,�0.01).23 For both of these groups, the effects dissipatedfter the interlocks were removed from the offenders’ehicles (HR�0.91, p�0.40, and HR�0.82, p�0.16,espectively).An evaluation42 of California’s interlock program didot separately estimate interlock effects before and afteremoval, but found net decreases in recidivism for a pe-iod that spanned the time that interlocks were installednd after they were removed. For this extended period,he hazard ratio for all offenders was 0.68 (p�0.05) andhat for second-time offenders was 0.59 (p�0.05). Fi-ally, an evaluation41 of a Swedish interlock program that
included an intensive alcohol treatment componentfound that the 171 participants had no re-arrests duringthe follow-up period relative to a recidivism rate of 4.4%per year among nonparticipants. However, these fındingsdo not include people who initially enrolled in the pro-gram but were expelled for failing to comply with thealcohol treatment plan.
Conclusions on effects of interlocks on recidi-vism. These fındings suggest that DWI offenders whohave ignition interlocks installed in their vehicles are atsubstantially lower risk for recidivism than those whohave had their licenses suspended either after beingdeemed ineligible for an interlock or deciding not tohave one installed. These fındings also suggest that theexperience of being enrolled in an ignition interlock
program by itself does not generally lead to long-term
ttotdwiabirvtls
368 Elder et al / Am J Prev Med 2011;40(3):362–376
changes in the propensity to drink and drive that lastbeyond the period of interlock installation.These fındings have to be considered in light of the
possibility that the observed differences in recidivismmay actually be due to preexisting differences betweenthe people who installed interlocks and those who didnot. Although this potential selection bias is an impor-tant factor to consider, two patterns in the resultssuggest that any such bias is likely to be small relativeto the overall effects of interlock installation. First, theresults from the single trial29 that randomized peopleo the interlock condition, and thus protected againsthis selection bias, were comparable to those from thether studies reviewed. Second, the fact that many ofhe included studies provide results for both the perioduring which interlocks were installed and after theyere removed allows estimation of the effectiveness ofnterlocks using the data from the interlock grouplone—thus avoiding any biasing effects of differencesetween people who did and did not have interlocksnstalled. The interlock groups’ dramatic increase inecidivism rates after the interlocks were removed pro-ides further evidence that the results of the compara-ive studies reflect true effects of participation in inter-ock programs and are not simply artifacts of groupelection.
Effects of Ignition Interlock Installation onMotor Vehicle CrashesThree of the included studies36,41,42 provided data on theeffects of participation in interlock programs on motorvehicle crashes. The results of the only study that found alower overall crash rate among the interlock group (0.0injury crashes per year relative to 0.6 per year for thecontrol group) were unreliable because of a very lowabsolute number of crashes studied.41 A study36 of theQuebec interlock program evaluated its effects on overallcrashes and on single-vehicle nighttime crashes (SVNCs;a proxy for alcohol-related crashes) during and after theperiod inwhich interlockswere installed. Rates of SVNCswere similar for fırst-time offenders with interlocksinstalled relative to those with suspended licenses(HR�1.05, p�0.85), and substantially, but nonsignifı-cantly, lower for repeat offenders (HR�0.46, p�0.14). Incontrast, total crashes were substantially higher for bothfırst-time offenders (HR�3.56, p�0.01) and repeat of-fenders (HR�2.16, p�0.01).The large differences in effect estimates for SVNCs
relative to total crashes for both fırst-time and repeatoffenders provide some evidence that interlocks pro-
tect against alcohol-related crashes, but that the instal-
lation of interlocks results in an increased overall crashrisk relative to that associated with having a suspendedlicense. Results from an evaluation42 of the Californiainterlock program support this conclusion. Partici-pants in the California interlock program had an 84%higher chance of being involved in a crash during thestudy period than the comparison group (p�0.05), andrepeat offenders had a 130% higher crash risk(p�0.05). However, the absolute crash rates for partic-ipants were similar to those for the general populationof California drivers.
ApplicabilityOne important issue to consider in any systematicreview is the potential applicability of the results tosituations in which the intervention is likely to beimplemented in the future. The studies included in thisreview primarily evaluated interlock programs that (1)were directed to “hardcore” drinking drivers, eitherrepeat offenders or fırst-time offenders who had highBACs at arrest (usually �0.15 g/dL), and (2) enrolled arelatively small subset of all DWI offenders. In con-trast, to maximize public health impact, interlock pro-grams will need to extend their reach to include abroader cross-section of offenders, and will need tofınd ways to ensure that a higher proportion of offend-ers actually have interlocks installed. It is reasonable toassume that interlocks will be effective at reducingrecidivism among the broader population of DWI of-fenders, with expected benefıts that are proportional totheir baseline rates of alcohol-impaired driving. None-theless, further research would be helpful to ensurethat interlocks remain effective as their reach isextended.One important caveat to the expectation that inter-
locks will be effective at reducing recidivism among thebroader offender population is that interlocks requiresubstantial administrative resources tomonitor partic-ipants. Thus, any major increase in program scope thatis not accompanied by an increase in administrativeresources may result in decreased effectiveness. Igni-tion interlock programs typically require offenders tobring their ignition interlock– equipped vehicle in forperiodic maintenance and checkup (typically every 30days). At these checkups, the data stored on the systemcan be downloaded and examined for signs of failedstart attempts, tampering, and circumvention. Thistype of intensive supervision and monitoring is anessential element of ignition interlock programs andmay play an important role in reducing recidivism
rates among program participants, both by helping to
www.ajpm-online.net
wdcfpaittcccmittmitcr
Elder et al / Am J Prev Med 2011;40(3):362–376 369
M
ensure compliance with program requirements and byeffectively communicating to offenders the need tochange their drinking and driving behaviors.
Other Benefits and HarmsThe primary goal of this review was to evaluate the effec-tiveness of ignition interlock programs for reducingalcohol-impaired driving and alcohol-related crashesamong people who have been arrested or convicted foralcohol-impaired driving (i.e., the specifıc deterrence offurther alcohol-impaired driving). These programs po-tentially have either benefıcial or harmful effects on thegeneral deterrence of alcohol-impaired driving in thebroader population (e.g., by heightening or lesseningthe perceived severity of the consequences of alcohol-impaired driving). Little is currently known about howignition interlock programs in general, or their specifıcfeatures, influence the general deterrence of alcohol-impaired driving.Participation in ignition interlock programs may be
associated with important benefıts and harms beyond theeffects on alcohol-impaired driving itself. These benefıtsand harms arise because of two other variables that inter-lock installation can be expected to affect—driving andalcohol consumption.
DrivingAlthough it is well known that people with suspendedlicenses often continue to drive,7 interlock participantsho can drive legally appear to make more trips andrive greater distances.32 The ability to drive legallyan have important benefıts to participants and theiramilies with respect to mobility. However, as in allopulations, increased exposure to driving is associ-ted with an increased crash risk. The small number ofncluded studies that examined the association be-ween interlock participation and crashes indicatedhat the interlock groups are involved inmore reportedrashes than comparison groups with suspended li-enses. However, they may not be involved in morerashes than the general driving population. Althoughore research on the association between interlock
nstallation and subsequent crashes would be valuable,he pattern of results in these studies is consistent withhe expected relationships from the current conceptualodel. Specifıcally, it appears that the increased driv-
ng exposure of interlock participants results in moreotal crashes than among those with suspended li-enses, but that there is no such increase for alcohol-
elated crashes.
arch 2011
Alcohol ConsumptionAlthough the reduced recidivism rates shown amonginterlock program participants is limited to the periodduring which the interlock is installed, substantial po-tential exists for synergistic use of interlocks with pro-grams to address offenders’ underlying alcohol depen-dence and abuse problems. By requiring participantsto choose between drinking and driving, interlocksprovide a consistent behavioral consequence (i.e., theinability to drive) that may discourage drinking onspecifıc occasions in the short term. Alcohol rehabili-tation or treatment services during the interlock pe-riod may be potentiated by this behavioral contin-gency. Longer-term alcohol recovery efforts can alsobe supported by integrating interlocks into a treatmentprogram as a source of objective data on compliance totreatment providers.
Economic EfficiencyIgnition interlock programs have several associated costs(e.g., program administration; leasing, installing, andmonitoring the device; auto insurance) and benefıts (e.g.,increased mobility for the offender, reduced alcohol-impaired driving) that may be important considerationsfor people making decisions about the structure of inter-lock programs and for offenders making decisions aboutparticipating in them. However, no studies of the costs oreconomic effıciency of ignition interlock programs thatmet the requirements for a Community Guide reviewwere identifıed.
Barriers to Intervention ImplementationFew barriers exist to the implementation of ignitioninterlock programs themselves, and 47 U.S. states havesuch programs. However, there are important barriersto devising interlock programs so that they enroll asuffıcient number of offenders to achieve the greatestpublic health impact. One important barrier to fulluptake of ignition interlocks among eligible offendersrelates to the lack of strong incentives for participationin interlock programs. There is almost always an explicitor implicit option to opt out of the interlock program(e.g., by claiming one does not have access to a vehicle),and the common alternative of license suspensionmay beviewed as less onerous than participation in an interlockprogram. This view is particularly likely if the offenderalready has been subjected to a period of license suspen-sion, subsequently has driven illegally, and has found thatthe risks of being caught and punished are acceptable.Reduction in the time period of pre-interlock licensesuspension, combined with improved enforcement of
and meaningful sanctions for driving while suspended,
e
370 Elder et al / Am J Prev Med 2011;40(3):362–376
ConclusionBased on the results of the studies identifıed in theCochrane review and those of the more recent studiesidentifıed in this review, there is strong evidence thatinterlocks are effective in reducing re-arrest rates whilethey are installed in offenders’ vehicles. The limitedavailable evidence from three studies suggests thatalcohol-related crashes decrease while interlocks areinstalled in vehicles. However, the potential for inter-lock programs to reduce alcohol-impaired driving andalcohol-related crashes is currently limited by the smallproportion of offenders who participate in the programsand the lack of a persistent benefıcial effect once the inter-lock is removed. More widespread and sustained use ofinterlocks among people arrested for DWI could have asubstantial impact on alcohol-related crashes.
DiscussionAlthough there is often an expectation that all DWI coun-termeasure programs, including ignition interlock pro-grams, will have a lasting impact on the drinking–drivingbehavior of offenders, recidivism rates remain at about 25%to30%.Asa temporary formof incapacitation imposed for afıxed period of time, interlock programs are able to reducerecidivismdramaticallywhile the interlock is in place.How-ever, theevidence indicates that it isunrealistic toexpect thatthe device will have persistent effects after removal in theabsence of additional program features. Unless interlocksare combined with interventions that address the underly-ing factors that contribute to recidivism—such as alcoholabuse and the lack of perceived alternatives to driving afterdrinking—it is likely that many users will continue to driveafter drinking once the device is removed.
Future Directions to Maximize theEffectiveness of Interlock ProgramsThe present review is based on studies of interlockprograms in various jurisdictions over the past 2 de-cades. These programs differ considerably in terms oftheir structure and operation. In fact, the only trulycommon feature of these programs is that DWI offend-ers had an interlock device installed for a given periodof time. At one level, this reflects on the robustness ofthe evidence for the effectiveness of ignition interlocks.At another level, it suggests opportunities to identifyand evaluate specifıc features of programs that showpromise for enhancing interlock effectiveness.This implies that the greatest need for research and
actions to improve the effectiveness of interlock pro- i
grams in reducing alcohol-impaired driving andalcohol-related crashes relates to the specifıcs of howprograms are implemented and operated. Some keyfeatures of interlock programs that could improvetheir effectiveness include increasing the time periodduring which the interlock is installed or making theremoval contingent on appropriate behaviors, usingthe interlock in conjunction with alcohol rehabilita-tion programs, increasing the number of participants,and improving protections against circumvention ofthe interlocks.The simplest approach to extending the benefıcial impact
of an interlock program is to extend the required period ofparticipation in the interlock program. The longer the de-vice is installed, the longer the period of protection fromrepeat offenses. Unfortunately, research studies to date pro-vide little guidance as to the ideal length of interlock pro-gram participation.An alternative approach is to eliminate fıxed periods
of interlock installation and implement performance-based criteria for removal based on objective indica-tors of participants’ performance during the period ofinstallation. In essence, before being eligible to havethe device removed, participants would have to dem-onstrate that they no longer need the interlock to pre-vent driving after drinking. This may require an ab-sence of any positive breath test result on the interlockdata recorder for a period of several months before theparticipant is eligible for release from the program anddevice removal. To reduce the possibility that offend-ers might simply park the vehicle for the fınal fewmonths to avoid having any positive breath test results,evidence of continued driving—either through odom-eter readings and/or a specifıed number of breathtests—would have to be included as part of the criteriafor removal. Alternatively, or additionally, offendersmight be required to submit proof that they success-fully had completed a treatment program or were nolonger using alcohol and, hence, were no longer at riskof impaired driving.A key element of efforts to extend the effect of inter-
lock programs may be to combine their use with par-ticipation in an alcohol rehabilitation program. Thiswould allow treatment providers to take advantage ofthe interlock recorder data to provide valuable infor-mation about alcohol use to inform treatment plan-ning andmodifıcation. One test of this approach foundthat the interlock provides useful information fortreatment specialists in promoting the recovery ofDWI offenders38 and identifıed several important ar-as for further developments that could enhance
mpact.
www.ajpm-online.net
Elder et al / Am J Prev Med 2011;40(3):362–376 371
M
The Swedish interlock program included in this re-view has a very strong alcohol rehabilitation compo-nent and may even be described as an alcohol treat-ment program that includes an interlock component.The focus of the program is abstinence from alcoholand living a sober lifestyle.41 The interlock is used toreinforce the goals of the program and to ensure that arelapse does not result in an impaired driving incident.This type of approach requires substantial cooperationand communication among interlock service provid-ers and rehabilitation professionals in the interest ofthe client.The ultimate consideration in evaluating the effec-
tiveness of interlock programs is their potential to havea meaningful effect at the level of the overall drivingpopulation. Even a highly effective intervention tar-geted toward individuals will not substantially im-prove public health unless it reaches an adequate num-ber of people. In the case of interlocks, many programscurrently have very low participation rates, and theirimpact on overall public health would undoubtedly beenhanced through substantially increased participa-tion rates. Taking steps to increase the number ofoffenders who are eligible for such programs and toincrease the proportion of eligible offenders who par-ticipate is necessary before overall reductions inalcohol-involved crashes will be realized.Typically, interlock programs have been targeted to
repeat and high-BAC offenders. However, fırst-timeDWI offenders more closely resemble repeat offendersthan they do non-offenders, and the results of thisreview suggest that interlocks are as effective with fırst-time DWI offenders as they are with repeat offenders.Thus, it would likely be a major boost to overall publicsafety to require fırst-time DWI offenders to partici-pate in an interlock program. Making participation insuch programs mandatory instead of at the discretionof judges, offenders, or both would also help to maxi-mize interlock use.However, to be truly effective, even mandatory pro-
grams require efforts to help ensure that the goals offull participation are met. These include minimizingthe availability of options for not participating, such asa lack of vehicle ownership; requiring installation atthe time of arrest rather than waiting until conviction;subsidizing participation for low-income offenders;and ensuring adequate follow-up to determinewhether offenders complied with the order to have aninterlock installed. Offering interlocks as an alterna-tive to a less attractive sanction also shows promise as ameans of increasing participation. For example, a pro-
gram that offered home confınement as an alternative
arch 2011
to interlock installation raised interlock participationrates to 62%,30 well above that of other programs.In addition, the effectiveness of ignition interlock pro-
grams potentially can be improved bymaking it more diffı-cult to circumvent the interlock device. The development ofimproved interlock hardware that is more resistant to cir-cumvention attempts, or detects them more rapidly, mayprovide incremental benefıts over existing hardware.Driveridentifıcation systems also show promise as a means to en-sure the driver actually provides the breath sample.However, greater efforts are required to develop effective
means of monitoring the most readily available method ofcircumvention—driving a non-interlock-equipped car.Monitoring the use of the vehicle through an analysisof the number of vehicle starts recorded on the inter-lock record or the mileage on the vehicle odometer iscurrently one of the only means of detecting suspecteddriving of other vehicles and appropriate use of that infor-mation is highlydependent on the effectiveness of the inter-lock monitoring agency. As the number of offenders us-ing interlocks increases, the incidence of driving vehiclesnot equipped with interlocks will likely become moreprominent, and novel means of deterrence and detectionwill be required.Perhaps the greatest need at present is for a uniform
set of guidelines or standards for ignition interlockprograms. Over the years, various groups have estab-lished technical standards for interlock devices thatserve to ensure that the hardware effectively prevents aperson who is impaired by alcohol from operating avehicle in which the interlock is installed. However,interlock programs require more than just the instal-lation of an interlock device in the vehicle of a con-victed DWI offender for a fıxed period of time. Theyrequire rules and regulations pertaining to eligibilityor requirements for program participation, length ofparticipation, extent of monitoring and reporting,agency responsible for monitoring, oversight of inter-lock providers, and consequences of repeated high-BAC readings or noncompliance with programrequirements.All these factors may play a role in determining the
overall success of the program. Aggregating these fea-tures into a set of standards or “guidelines for bestpractices” based on existing research fındings wouldfacilitate harmonization and enhance the overall suc-cess of interlock programs. An initial set of such guide-lines for best practice has been prepared for Canadianinterlock programs.44
Finally, it is important to establish and maintain afocus on the primary goals for an ignition interlockprogram. First and foremost, these goals must ac-
knowledge the role of the interlock as an instrument of
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
372 Elder et al / Am J Prev Med 2011;40(3):362–376
incapacitation, a means to prevent a convicted DWIoffender from operating a vehicle while under the in-fluence of alcohol. Building a comprehensive rehabili-tation program for DWI offenders that incorporatesinterlocks is a worthy endeavor and one that has tre-mendous potential for a substantial overall impact onalcohol-impaired driving.
The fındings and conclusions in this report are those of theauthors and do not necessarily represent the views of the CDC.No fınancialdisclosureswere reportedby theauthorsof thispaper.
References1. Hedlund J, Fell J. Persistent drinking drivers in the U.S. Des Plaines
IL: Association for the Advancement of Automotive Medicine,1995:1–12.
2. Simpson HM. Who is the persistent drinking driver? Part II: Canadaand elsewhere. Transp Res Circ 1995;437:21–35.
3. Vingilis E, Stoduto G, Macartney-Filgate MS, Liban CB, McLellan BA.Psychosocial characteristics of alcohol-involved and nonalcohol-involved seriously injured drivers. Accid Anal Prev 1994;26:195–206.
4. Voas RB. Evaluation of jail as a penalty for drunken driving. AlcoholDrugs Driving 1986;2:47–70.
5. Peck RC, Sadler DD, Perrine MW. The comparative effectiveness ofalcohol rehabilitation and licensing control actions for drunk driv-ing offenders: a review of the literature. Alcohol Drugs Driving1985;1:15–40.
6. Nichols JL, Ross H. The effectiveness of legal sanctions in dealing withdrinking drivers. Alcohol Drugs Driving 1990;6(2):33–60.
7. Ross HL, Gonzales F. The effects of license revocation on drunk-driving offenders. Albuquerque NM: University of New Mexico, De-partment of Sociology, 1987.
8. McCartt AT,Geary LL,NissenWJ.Observational study of the extent ofdriving while suspended for alcohol-impaired driving (DOT HS 809491). Washington DC: U.S. Department of Transportation, NationalHighway Traffıc Safety Administration, 2002.
9. Tashima HN, Helander CJ. 1999 annual report of the California DUIManagement Information System (CAL-DMV-RSS-99-179). Sacra-mento CA: California Department of Motor Vehicles, Research andDevelopment Section, 1999.
0. DeYoung DJ. An evaluation of the specifıc deterrent effects of vehicleimpoundment on suspended, revoked, and unlicensed drivers in Cali-fornia. Accid Anal Prev 1999;31:45–53.
1. Voas RB, Tippetts AS, Taylor EP. Temporary vehicle immobiliza-tion: evaluation of a program in Ohio. Accid Anal Prev1997;29:635–42.
2. Voas RB, DeYoung DJ. Vehicle action: effective policy for control-ling drunk and other high-risk drivers? Accid Anal Prev2002;34:263–70.
3. Peck R, Voas R. Forfeiture programs in California: why so few? J SafetyRes 2002;33:245–58.
4. VoasRB.Cars that drunks can’t drive. Paper presented atAnnualMeetingof the Human Factors Society; 1970 October; San Francisco CA.
5. Collier DW. Second generation interlocks lead to improved programeffıciencies (TRB ID Number: CF076). Paper presented at the 73rdAnnual TRB (Transportation Research Board) Meeting; 1994 January9–13; Washington DC.
6. Collier DW, Comeau FJE, Marples IR. Experience in Alberta withhighly sophisticated anti-circumvention features in a fuel cell basedignition interlock. Adelaide, Australia: NHMRC Road Accident
Research Unit, 1995.
7. Voas RB, Marques PR, Tippetts AS, Beirness D. Circumventing thealcohol safety interlock: the effect of the availability of a noninterlockvehicle. Stockholm, Sweden: ICADTS, 2000:1133–42.
8. Marques PR. Alcohol ignition interlocks—past, present and future. In:Nickel W, Meinhard G, Born I, eds. Fit to drive. 4th InternationalTraffıc Expert Congress. Tallinn 2009:66–72.
9. Voas RB, Blackman KO, Tippetts AS, Marques PR. Motivating DUIoffenders to install interlocks: avoiding jail as an incentive. AlcoholClin Exp Res 2001;25:102A.
0. Baker EA, Beck K. Ignition interlocks for DWI offenders: a useful tool.Alcohol Drugs Driving 1991;7(2):107–15.
1. DeYoung DJ, Tashima HN, Masten AS. An evaluation of the effective-ness of ignition interlock in California: report to the legislature of theState ofCalifornia (CAL-DMV-RSS-04-210/AL0357). SacramentoCA:California Department of Motor Vehicles, 2004.
2. DeYoungDJ.Anevaluationof the implementationof ignition interlock inCalifornia (CAL-DMV-RSS-02-195/AL9814). Sacramento CA: Offıce ofTraffıc Safety, California Department ofMotor Vehicles, 2002.
3. Roth R, Marques P, Voas R. New Mexico ignition interlock: laws,regulations, utilization, effectiveness, cost-effectiveness and fairness.Presentation to the 8th Annual Ignition Interlock Symposium, August26–27, 2007, Seattle WA.
4. Voas R, Marques P. Barriers to interlock implementation. Traffıc InjPrev 2003;4:177–82.
5. McKnight AJ, Voas RB. The effect of license suspension upon DWIrecidivism. Alcohol Drugs Driving 1991;7:43–54.
6. Voas RB, Roth J, Marques PR. The hard suspension barrier: does NewMexico’s interlock licensing law solve the problem? In: Marques PR,ed. Alcohol ignition interlock devices volume II: research, policy, andprogram status 2005. Oosterhout, the Netherlands: InternationalCouncil on Alcohol, Drugs and Traffıc Safety, 2005:62–73.
8. Briss PA, Zaza S, PappaioanouM, et al. Developing an evidence-basedGuide to Community Preventive Services—methods. Am J Prev Med2000;18(1S):35–43.
9. Beck KH, Rauch WJ, Baker EA, Williams AF. Effects of ignition interlocklicense restrictions on drivers with multiple alcohol offences: a randomizedtrial inMaryland.AmJPublicHealth 1999;89(11):1696–700.
0. Voas RB, Blackman KO, Tippets AS, Marques PR. Evaluation of aprogram to motivate impaired driving offenders to install ignitioninterlocks. Accid Anal Prev 2002;34:449–55.
1. EMT Group. Evaluation of the California ignition interlock pilot pro-gram for DUI offenders. Sacramento CA: California Department ofAlcohol and Drug Programs, 1990.
2. Morse BJ, Elliott DS. Effects of ignition interlock devices on DUIrecidivism: Findings from a longitudinal study in Hamilton County,Ohio. Crime Delinq 1992;38(2):131–57.
3. Jones B. The effectiveness of Oregon’s ignition interlock program.Proceedings of the 12th International Conference on Alcohol, Drugs,and Traffıc Safety, ICADTS-T 92; 1992 Sep; Cologne, Germany. Co-logne: Verlag TUV Rheinland, 1993.
4. Popkin CL, Stewart JR, Beckmeyer J, Martell C. An evaluation of theeffectiveness of interlock systems in preventing DWI recidivismamong second-time DWI offenders. Cologne: Verlag TUV Rhein-land, 1993.
6. Vezina L. The Quebec alcohol ignition interlock program: impact onrecidivism and crashes. In: MayhewD, Dussault C, eds. Proceedings ofAlcohol, Drugs and Traffıc Safety—T 2002: 16th Annual Conferenceon Alcohol, Drugs and Traffıc Safety; 2002 Aug 4–9. Vol. 1. QuebecCity: Societe de L’assurance Automobile du Quebec, 97–104.
7. Tippetts AS, Voas RB. The effectiveness of theWest Virginia interlock
program. J Traffıc Med 1998;26:19–24.
www.ajpm-online.net
Elder et al / Am J Prev Med 2011;40(3):362–376 373
M
38. Voas RB,Marques PR, Tippetts AS, Beirness DJ. The Alberta InterlockProgram: the evaluation of a province-wide program on DUI recidi-vism. Addiction 1999;94(12):1849–59.
39. Marine W. High-tech solutions to drinking and driving: evaluationof a statewide, voluntary alcohol ignition interlock program. Finalgrant report. Denver CO: University of Colorado Health SciencesCenter, 2001.
40. MarineW.High-tech solutions to drinking and driving: evaluation of astatewide, voluntary alcohol ignition interlock program. Final grantreport. RWJF IDNumber 028805. Denver CO: University of ColoradoHealth Sciences Center, 2001.
41. Bjerre B. Primary and secondary prevention of drinking and drivingby the use of Alcolock device and program: the Swedish experience.In: Marques PR, ed. Alcohol ignition interlock devices, Volume II:
arch 2011
research, policy, and program status 2005. International Council onAlcohol, Drugs and Traffıc Safety (ICADTS), Oosterhout, TheNetherlands, 2005:11–24.
42. DeYoung DJ, Tashima HN, Masten SV. An evaluation of the effec-tiveness of ignition interlock in California. In: Marques PR, ed.Alcohol ignition interlock devices, Volume II: research, policy, andprogram status 2005. International Council on Alcohol, Drugs andTraffıc Safety (ICADTS), Oosterhout, The Netherlands, 2005:42–51.
43. Roth R, Voas R, Marques P. Mandating interlocks for fully revokedoffenders: the New Mexico experience. Traffıc Inj Prev2007;8:20–25.
44. Beirness DJ. Alcohol ignition interlock program standards for Canada.Ottawa: Transport Canada and the Canadian Council of Motor Trans-port Administrators, 2007.
Appendix A. Evidence table of study details, program details, and results for studies included in this review
Study (Study period)Study designEvaluation setting Study details
Intervention n�273Comparison n�273Comparison group matched
on (1) problem drinkerclassification; (2)number of DUI arrests;and (3) number of non-DUI alcohol/drugarrests
DMVCourt discretion for offenders with
(1) BAC�0.20, (2) BAC testrefused, or (3) repeat offenders;participation optional
12–30 months40.5% of eligible offenders
Re-arrest during interlock periodAll participants: RR�0.33
(0.15, 0.73)People who opted for interlock
installation drove moremiles than those who didnot (e.g., 42% vs 30%drove more than 200miles/week)
30 months
Jones (1993)33
(1 year; 1988–1989)Prospective cohort
with concurrentcomparison group
Oregon
Intervention n�648Comparison n�1541Comparison group consisted
of drivers in comparisoncounties who reinstatedtheir licenses
DMVOptional for offenders who have
completed 1–3 years of “hard”license suspension (with noadditional suspensions during thatperiod)
6 months (in lieu of 6 months’additional license suspension)
18% of eligible offenders
Re-arrest during interlock periodRepeat offenders: RR�0.60
(0.35, 1.04)Re-arrest following interlock
periodRepeat offenders: RR�0.94 (0.73,
1.20)Judges tended to select more
serious, habitual offendersfor interlock program;offenders who acceptedinterlocks were more likelyto be white, have higherincomes, and havemultiple prior DUIs
M��21 months (6with interlocksinstalled)
Popkin (1993)34
(1/1986–3/1992)Prospective cohort with
concurrentcomparison group
North Carolina
Intervention n�407Comparison n�916Comparison group consisted
of drivers who weregranted a conditionallicense that did notrequire interlockinstallation
DMVOptional for offenders who have
completed 2 years of “hard”license suspension
24 months (in lieu of 24 months’additional license suspension)
1.8% of eligible offenders
Re-arrest during interlock periodRepeat offenders: RR�0.38
(0.20, 0.71)Re-arrest following interlock
periodRepeat offenders: RR�1.07
(0.53, 2.18)
24 months
Raub (2003)35
(7/1991–6/2000)Before/after studyIllinois
Intervention n�1560Comparison n�1384Comparison group consisted
of drivers who receivedrestricted drivingpermits (RDPs) in the 3years prior to theinterlock program (i.e.,7/91–6/94)
DMVMandatory for offenders who applied
for RDPs following a minimum 180-day suspension period
12 months�14% of eligible drivers
Re-arrest during interlock periodRepeat offenders: RR�0.19
(0.12, 0.30)Re-arrest in 2 years following
interlock periodRepeat offenders: RR�0.52
(0.41, 0.65)b
Drivers in interlock group wereolder than those incomparison group (meanage of 38.7 vs 37.5 years,p�0.05)
36 months
Vezina (2002)36
(12/1997–1/2001)Prospective cohort with
concurrentcomparison group
Quebec
InterventionFirst-time offenders:
n�8846Repeat: n�1050
ComparisonFirst-time offenders:
n�25,559Repeat: n�7108
Comparison group consistedof drivers who did notparticipate in theinterlock program
DMVOptional9 months (first-time offenders) or 18
months (repeat offenders)26% of first-time offenders; 13% of
repeat offenders
Re-arrest during interlock periodFirst-time offenders: RR�0.20
(0.14, 0.29)Repeat offenders: RR�0.34
(0.22, 0.53)Re-arrest following interlock
periodFirst-time offenders: RR�1.37
(1.21, 1.56)Repeat offenders: RR�1.93
(1.02, 3.66)Single-vehicle nighttime crashes
during interlock period
36 months
(continued on next page)
www.ajpm-online.net
Elder et al / Am J Prev Med 2011;40(3):362–376 375
M
Appendix A. (continued)
Study (Study period)Study designEvaluation setting Study details
Re-arrest during interlock periodRepeat offenders: HR�0.35
(p�0.01)Re-arrest following interlock periodRepeat offenders: HR�0.91
(p�0.40)
66 months
Roth (2007)23
(1/2003–12/2005)Prospective cohort with
concurrentcomparison group
New Mexico
Intervention n�1461Comparison n�17,562Comparison group consisted
of all first-timeoffenders withoutinterlocks
CourtsMandatory for offenders with high
BAC (�0.16 g/dL) or in injurycrashes
Mean installation period�197 days8.8% of offenders with BAC �0.16
g/dL
Re-arrest during interlock periodFirst-time offenders: HR�0.39
(p�0.01)Re-arrest following interlock periodFirst-time offenders: HR�0.82
(p�0.16)Re-arrest during study period (during-
and post-interlock installation)First-time offenders: HR�0.61
(p�0.61)Interlock group tended to be older
(35.7 vs 31.7 years), withmore men and high-BACoffenders
36 months
aResults differ from those in Cochrane review (HRs reported, rather than RRs).b
Results differ from those in Cochrane review (based on longer follow-up time).BAC, blood alcohol content; DMV, Department of Motor Vehicles; DUI, driving under the influence; HR, hazard ratio; RDP, restricted driving permit; RR, risk ratio
